Pandas interview question: find the bugs in an average order value notebook

EasyCode review

The colleague’s AOV doesn’t equal revenue divided by the number of orders. For each platform, find September 2026 orders, their revenue and AOV.

• aov = revenue_usd / orders

# AOV by platform, September
sept = orders[(orders['status'] == 'delivered')
              & (orders['created_at'] >= '2026-09-01')
              & (orders['created_at'] < '2026-10-01')].copy()
sept['day'] = sept['created_at'].str[:10]

# revenue, orders and AOV by day
daily = (sept.groupby(['platform', 'day'])
             .agg(revenue_usd=('amount_usd', 'sum'),
                  orders=('order_id', 'count'))
             .reset_index())
daily['aov'] = daily['revenue_usd'] / daily['orders']

# platform totals
result = (daily.groupby('platform')
               .agg(orders=('orders', 'count'),
                    revenue_usd=('revenue_usd', 'sum'),
                    aov=('aov', 'mean'))
               .reset_index())

Example

iOS: $120 over 4 orders, so $30. Android: $48 over 2 orders, so $24.

Output columns

platform

the order’s platform (orders.platform)

Skills it tests

• Reading pandas code
• Averages and ratios

More interview questions

• Pandas: Rolling 7-day revenueEasy · Rolling aggregates
• Pandas: Cart-to-order conversionMedium · Funnels and conversion
• SQL: Second-highest price in each groupEasy · Nth-highest value
• SQL: Top 3 categories in each cityMedium · Top-N per group
• SQL: Day-1 retention by channelMedium · Retention

All questions